slow traveling and rethinking the human-machine interface | formerly engineer at @MojoVisionInc, @Microsoft, @UWdub | DMs opens
Seattle, WA
Joined June 2009
- Tweets715
- Following1K
- Followers658
- Likes1.2K
I didn’t expect my first “can’t-put-it-down” book of the year to be… a nonfiction about the 1893 Chicago World’s Fair and a serial killer. But here we are. 😅
I picked up The Devil in the White City by Erik Larson, and it’s wild in the best way. Two threads running side by side: Daniel Burnham, the architect behind the Fair and a serial killer who used the chaos and crowds to lure victims.
What’s sticking with me (so far) is this Burnham line: “Make no little plans; they have no magic to stir men’s blood and probably will not themselves be realized.”
It immediately made me think of Sam Altman’s take: “The most counterintuitive secret about startups is that it’s often easier to succeed with a hard startup than an easy one.”
Different eras, same idea: big, audacious goals create energy. They attract talent. They pull you through inevitable mess. They give people a reason to care.
Agent ecosystem needs to evolve beyond text -> into UI.
To get there, it looks like we’re converging on two strategies. Here are my notes from a third-party perspective:
MCP/ChatGPT Apps
- you ship a web UI component that the host embeds in an iframe
- UI controlled by you; can be more rich and interactive
- you make sure it works on cross-platform surfaces (mobile, web, etc)
- now under Agentic AI Foundation, Linux Foundation
A2UI by Google
- your agent emits a declarative UI spec (data, not executable code) for the host to render
- more native UI experience for user
- you don't have to worry about cross-platform differences
- A2A is under Linux Foundation, A2UI could follow a similar path
It's not one or the other, a host can support both. It’ll be interesting to see how adoption plays out in 2026
Airbnb MCP server—now 250+ stars on github
Add it to Claude Desktop and you can simply ask:
“Find me top Airbnbs in Seattle for the first week of Sep, under $250/night, near a gym and grocery store.”
Learning to code is more important NOW than ever.
To vibe code is to command an army of tireless bots that can build whatever you want.
But you gotta be a soldier first before being a good commander.
Such an amazing time to be alive!
Or just install openbnb.org extension once and the direct website will be ready for you on the listing page
Aniket retweeted
The most interesting startups to be built around AI are ahead of its current capability, UIs that seem quite different than a prompt box, and serve a new set of customers entirely. I suspect they aren’t the easy-to-explain existing markets with well established incumbents.
Woah! Already at 17,500+ downloads for the Airbnb MCP server in few weeks!
From youtube tutorials to blogs, its going places.
Let me know if you need help setup; imo its the best way to find Airbnbs using AI.
This quoted post is unavailable.
Thanks for all your support in the Product Hunt launch.
OpenBnB browser extension was #3 product of the day!
In last two days, we have added 500+ active users along with tons of Airbnb hosts submitting their direct booking website.
We are saving hundreds of travelers ~$240 on their stays when they directly book from 160,000+ listings on OpenBnB 🚀
Airbnb MCP server with Google DeepMind
Here is a fully Agentic example on how to use MCP servers with @GoogleDeepMind Gemini. It shows how to use the Airbnb MCP to find rooms in Paris this weekend for a booking.
> Define MCP server
> Convert MCP tools to Gemini
> Initial Request
> Agentic Loop for automatic tool calling
> Final Response
code ⬇️
